1. Field of the Invention
The present invention relates to a method of verifying correct operation of a tester for semiconductor devices, and to a method for creating a test standard device for use with such a tester.
2. Description of the Related Art
After manufacture and prior to shipment, semiconductor devices are typically subjected to a number of tests to verify their performance and reliability. Among these is an electrical performance test.
The testing apparatus conducts electrical performance tests by measuring certain electrical characteristics of the semiconductor devices and sorts them into a number of "bin" categories according to the test results. For example, the semiconductor devices classified as "BIN1" (i.e., bin category 1) are those that pass, or fall within, the electrical performance specification, in respect to all test items, and devices classified as BIN7 (i.e., bin category 7) are those that fail, or are outside of specification, in respect to all test items. BIN2 through BIN6 are reserved for the devices that are within specification in respect to certain test items and out of specification with respect to others. Of course, the bin numbering system used may be different than that described above, depending on the particular manufacturer and devices being tested. However, it is common within the industry that BIN1 represents devices that fall within specification on all test items.
For assuring the fidelity of semiconductor device tests, it is necessary to verify that the apparatus used to test the devices, i.e., the tester itself, is functioning correctly. In a conventional test method, when the tester sorts a "standard device," i.e., one that has previously been determined to be within specification on all test items, into BIN1, the tester is deemed to be functioning properly. On the other hand, if the tester does not sort the standard device into BIN1, the tester is suspected to be malfunctioning. That is, if the tester sorts the standard device into any BIN, e.g., BINX, other than BIN1, it is presumed that the tester is malfunctioning, at least with respect to the particular test item(s) related to BINX.
Also, in accordance with the conventional method, the test engineer initially selects a test standard device from a group of candidate devices. The engineer must test each of the candidate devices manually, then select a standard device from the group by means of a laborious statistical analysis of the test results.
In the conventional test method, it is presumed that the tester is malfunctioning, at least with respect to a particular test item, when the number of defective devices sorted by that tester to a specific BIN is higher than the average number of defective devices in that BIN. However, that is not always the case, as it is possible that the tested devices are actually out of specification in regard to the specific test item. In addition, when replacement of a standard device is necessary, the test engineer must manually select the standard device by application of his/her subjective judgment to voluminous test results, as described above, and thus, the creation of a precise test standard device can be a very cumbersome and laborious task.